LIBOR rigging

The Jolly Contrarian Law Reports Our own, snippy, in-house court reporting service. →Go uptown Dramatic look gopher goes to the British Bankers’ Association (von Sachsen-Rampton, 2024) Editorial Board of the JCLR: Managing Editor: Lord Justice Cocklecarrot M.R. · General Editor: Sir Jerrold Baxter-Morley, K.C. · Principle witness: Mrs. Pinterman Common law | Litigation | Contract | Tort | Click ᐅ to expand: Case notes Tell me more Sign up for our newsletter — or just get in touch: for ½ a weekly 🍺 you get to consult JC. Ask about it here.

“If the law supposes that,” said Mr. Bumble, squeezing his hat emphatically in both hands, “the law is a ass—a idiot. If that’s the eye of the law, the law is a bachelor; and the worst I wish the law is that his eye may be opened by experience—by experience.”

— Charles Dickens, Oliver Twist

LIBOR: deep background

Banks have structural interest rate risk

The basic model of a bank is to borrow, short-term, at a low rate, and lend, long-term, at a high rate. Generally, banks calculate interest on deposits, by which they borrow, at a floating rate and on term loans, by which they lend, at fixed rates.

There is a straightforward reason for this: call deposits don’t have a term; they can be withdrawn at any time. All you can do is apply a prevailing daily rate. On the other hand most people borrow for a fixed term and want certainty on how much interest they must pay, so prefer fixed interest.

Since banks borrow in floating and lend in fixed, they have “structural interest rate risk”. It is a natural function of how banks work. They want floating rates to be low, and to move lower. If they don’t manage this risk, things can get funky, fast. Just ask Silicon Valley Bank.

So knowing what that floating rate is, and managing it, is an important risk management function for the bank. A risk well managed is called a “return”. The floating rate is different from the central bank’s base rate, and moves daily in response to market conditions.

Where does this “floating rate” come from, then?

In the good old days, each bank worked out its own floating rates based on its own models, funding costs and market positioning. This process was neither transparent nor standardised. Rates could vary significantly between similar banks. As long as interest rates were not tradable instruments, this did not much matter to banks: they just told their customers what the floating rate was each day, and that was that.

In the early nineteen eighties, some bright sparks at Salomon Brothers figured out how to make interest rates into a tradable instrument. To standardise that instrument, the banks realised they would need a common way of describing how their interest rates change through time. A “benchmark”.

Chess club

Enter the the British Bankers’ Association. This was just the sleepy, city-grandees-in-a-smoke-filled-gentlemen’s-club-in-Threadneedle-Street of your imagination. It began to compile what it called the “London Interbank Offered Rate” — “LIBOR”. This was to be an objective distillation of all the major banks’ borrowing rates.

The method the BBA chose to compile it was simple: it invited 18 major banks to literally, phone in what they believed they could borrow in various currencies and maturities in the market each day. The BBA would then compile the submissions, “trim” off the top and bottom four, average the rest and publish a set of daily LIBOR rates for each currency and maturity, before toddling off for a liquid lunch at the Garrick and their regular three o’clock tee time at Wentworth.

You get the picture.

With LIBOR published, the banks could then set their rates for call deposits, calculate suitable fixed rates for new term loans, and more importantly trade standardised interest rate instruments by reference to the new LIBOR “benchmark”.

Happy, unadventurous stuff, carried out by happy, unadventurous people. Look: we don’t want to run the interest rate-setting crowd down, but before 2007, the LIBOR rate setting process was like the after-school chess club: snoresville. All the cool kids were out shagging, smoking weed and shorting structured credit. None of the hepcats paid much attention to LIBOR.

Now. It is one of JC’s axioms of financial scandal that calumny happens where you least expect it. This is because success in financial services is in large part about “edge”, and you generally only find an edge where no-one else is looking for it.

The cool kids

Tom Hayes was a cool kid (metaphorically: literally he has been described as “socially awkward”) but he hung out in the chess club. He, and a bunch of other groovers, found some edge there, where no one was looking for it. No one bothered them and they didn’t do a lot of harm — not, at least, that anyone has been since able to point to. But they sent each other lots of embarrassing emails.

In any case, they made an effort to submit LIBOR rates that suited their derivatives trading positions and not, necessarily, their banks’ structural interest rate positions.

That this all came to light as a result of the unrelated “lowballing” scandal, after which lots of people began looking very hard at LIBOR, and not liking what they saw.

Another one of JC’s axioms: if you like sausages, don’t work in a smallgoods factory.

As per the “basic banking model”, to manage its structural interest rate risk, a bank generally would want LIBOR to be low. But deposits are not the only show in town — there are other exposures to the interest rate market: notably, the new tradable instruments: interest rate swaps.

Interest rate swaps

In an interest rate swap, the bank “swaps” interest rates with individual counterparties: it might, for an agreed period, pay one counterparty a fixed rate and receive from it a floating rate; with another it might pay floating and receive fixed.

Before the advent of swaps, the only way of getting exposure to interest rates was by borrowing and lending principal. This required a lot of money down. Interest rate swaps got popular, fast. There are now trillions of dollars in notional interest rate swaps outstanding on any day.

Unlike basic banking, there is no structural bias to swap trading. If a bank swaps a five-year fixed rate for a five-year floating rate, and LIBOR then goes up, by definition the bank profits: the “present value” of its incoming floating rate will increase while the present value of its outgoing fixed rate stays the same. The dealer is therefore “in-the-money”. If it swapped floating for fixed in the same case, it would book a corresponding loss.

While banks try to balance their books so their portfolio of customer swaps offset each other as far as possible, how they “position” the book might help manage the bank’s structural interest rate risk.

Under the “basic banking model”, a bank will always be “axed” for floating rates to be as low as possible. You would expect a basic bank’s LIBOR submissions to reflect that. But a swap trader who is “long” floating rates will wish floating rates to go higher.

This prospect, we venture, was not wildly present in the minds of the Sir Bufton Tuftons who formulated the LIBOR rules that defined how submitting banks should choose the rates they submit each day.

The question arose later, even though it did not arise then: when submitting a rate, what account, if any, may a bank take of its own derivatives trading book?

The LIBOR Definition

The BBA’s guidance came in the form of “Instructions to BBA LIBOR Contributor Banks”. The critical part of these — what the court called the “LIBOR Definition” — ran as follows:

“An individual BBA LIBOR Contributor Panel Bank will contribute the rate at which it could borrow funds, were it to do so by asking for and then accepting inter-bank offers in reasonable market size just prior to 1100.”

On any day there will be a range of rates at which a bank could borrow. These might be firm offers from other lenders, good faith estimates or model outputs. There is an excellent subjunctive in there, by the way: “were it to do so” implies that that a submitting bank need not actually do so.

Say the range of available rates a bank sees on a given day is between 2.50% and 2.53%. Which of these is “the rate at which it could borrow funds”? You can only choose one.

Setting aside for a moment compliance with the LIBOR Definition, the possible avenues open to a bank in submitting a rate are:

Pick an “available” rate: Choose one of the rates from the range, as above.

Manufacture a blended rate from the range: Contrive some artificial rate from within that range, reflecting a median, a weighted average, or some such thing.

Make one up: Submit a rate that did not fall within the estimated range, whether lower or higher.

“Making one up” plainly falls outside the scope of the LIBOR Definition. “Making a blended rate” does not quite conform to its text, but perhaps captures its spirit.

To an uncomplicated reading, “picking one of the available rates” seems to fall squarely within the LIBOR Definition. This was a rate at which the bank could borrow funds.

This is what Hayes did. The complication is that he actively selected the available rate that best suited his or, in some cases, competitors’ derivative trading positions. That is, he was guided by his own commercial interests, and not the “structural” interests of a hypothetical basic bank.

This is the crux of the case: was this ulterior motive dishonest in light of the “proper basis for the submission of those rates”? The Crown alleged it was.

The criminal charges

Okay, so a picture is emerging. During the 1980s the “interest rate” transformed from being the intractable time cost of borrowing money — for lenders, a secondary risk to the repayment of principal — to a tradable asset class of its own, thanks to the emergence of interest rate swaps.

This is a profound conceptual shift. Like the transformation from analogue to digital: you could lift interest rate information from the substrate in which, hitherto it was buried, in the same way that the information content of a book was suddenly abstracted from the paper it was printed on. But there was a difference: digital information, is logically prior to the substrate in which it is articulatd. An interest rate is not logically prior to a loan. It is a consequence of a loan. An interest rate cashflow implies a loan.

In any case, conceptually free of that mortal weight of principal, and enabled by the rise of computer modelling — the coincidence of the digital and derivative revolutions was, ah, no coincidence — financial modellers were able to create and then , with their new tools, hedge, all kinds of funky new nterest rate products. Collars, caps, floors, enhanced dual rates , knock ins, knock outs. There is another story here — in JC’s view far more outrageous — about how commercial bankers used these tools to shift interest rate risks onto small businesses, but that will have to wait.

In the meantime, “what the LIBOR rate might be used for” quickly changed. Sleepy old deposits rates and mortgages were but a small part of it.

Banks across the street engaged their derivatives trading teams in the LIBOR setting process. Tom Hayes was one such derivatives trader. He traded Yen LIBOR in Tokyo. Though a young man at the time of the allegations, Hayes was valued by a succession of bank employers — for a time — for his great connections around the LIBOR setting market.

As public interest focussed in on the LIBOR submission process in the wake of the LIBOR lowballing incident sentiment — both inside and outside his employer ls — turned sharply against him.

“A conspiracy to defraud”

Hayes was indicted on the ancient common law offence of “conspiracy to defraud”. Criminal law minutiae, perhaps, but he was not charged under the Fraud Act 2006. That Act followed a Law Commission survey of the ancient criminal law of fraud which recommended abolishing common law conspiracy to defraud, because of its “potential to catch behaviour that should not be criminal”.

Though the government accepted the general thrust of the Law Commission’s recommendations, it “decided to retain [common law conspiracy to defraud] for the meantime, but accepted the case for considering repeal in the longer term.”

In any case, common law conspiracy to defraud was not abolished, still hasn’t been, and that is what Tom Hayes was charged with.

Being a common law offence, its ingredients are not sharply delineated — this in itself is a policy reason to to prefer statutory crimes, but anyway — though they seem to be along the following lines: an agreement between persons intending to defraud someone by doing something dishonest with a likelihood of resulting loss, even if no loss eventually arises.

The crux: was Hayes dishonest when he submitted his LIBOR rates?

That, the court thought, came down to whether he “deliberately disregarded the “proper basis” for the submission of those rates”.

The court did not dwell on what the “LIBOR Definition” meant — there’s not much to dwell on — but rather asked whether Hayes’ intention when choosing the rate to submit reflected “the bank’s genuine perception of its borrowing rate”, instructing the jury as follows:

“1. Did Mr Hayes agree with any individual as named in the counts, to procure the making of a submission by a bank of a rate which was not that bank’s genuine perception of its borrowing rate for the tenor in question in accordance with the LIBOR definition but was a rate which was intended to advantage Mr Hayes’s trading?

If the answer is No, Mr Hayes is not guilty on that Count. If the answer is Yes, proceed to Question 2.

2. Was what Mr Hayes did dishonest by the ordinary standards of reasonable and honest people?

If the answer is No, Mr Hayes is not guilty on that Count. If the answer is Yes, proceed to Question 3.

3. Did Mr Hayes appreciate that what he was doing was dishonest by those standards?

If the answer is No, Mr Hayes is not guilty on that Count. If the answer is Yes, Mr Hayes is guilty on that Count.”

The jury answered, “yes” to all three questions. Hayes was sent to prison for 14 years.

Tom Hayes was not the only one. In total, thirty-seven traders were prosecuted in London and New York for interest rate benchmark manipulation. Of these, nineteen were convicted and nine imprisoned.

At the time, there was plenty of righteous dudgeon about LIBOR rigging. None of it favoured Tom Hayes or his fellow inmates, who fitted a popular narrative. Even their employers affected a tone of wounded indignance that their reputations can have been so rudely traduced. It is a familiar stance.

But if little old ladies make bad law, then what about young investment bankers?

Meanwhile, in Gotham City

The travails of other LIBOR submitters are interesting because of the sheer scale of the ostensible “criminal enterprise” — we’ll come to that — but also because two of those convicted in the United States appealed in 2022. Their appeal focused on what the LIBOR Definition actually meant.

In United States v Connolly and Black the United States Court of Appeals for the Second Circuit found construing the LIBOR Definition to be a question of fact: filtered through the prisms of grammar, usage, and context, an upon which evidence of industry practice from subject matter experts would have a bearing.

The question of law — were the submitters dishonest? — depended a great deal on the LIBOR Definition as to which the US court took a literal reading:

The precise hypothetical question to which the LIBOR submitters were responding was at what interest rate “could” DB borrow a typical amount of cash if it were to seek interbank offers and were to accept. If the rate submitted is one that the bank could request, be offered, and accept, the submission, irrespective of its motivation, would not be false.

This led the US court to conclude that picking from a range of available rates, however motivated, could not be fraudulent.

“Here, the government failed to show that trader-induced LIBOR submissions did not reflect rates at which DB could have borrowed. If the submissions did reflect rates at which DB could have borrowed, they complied with the BBA LIBOR Instruction, and the LIBOR submissions were not false.”

It was within the rules. Connolly and Black were acquitted.

Buoyed by the outcome in New York, Tom Hayes persuaded the UK Criminal Cases Review Commission to refer his case back to the Court of Appeal — which, significantly, had aleady heard and rejected his appeal once — for reconsideration.

The Court of Appeal handed down its decision in March 2024.

The Hayes appeal

The Court of Appeal considered first that question of legal methodology — whose job was it to determine what the LIBOR Definition meant — and came to a different conclusion.

Under English law, contractual interpretation is a matter of law, to be resolved by the judge. Evidence of market practice, or the subjective belief of submitters, does not enter into it.

The Court of Appeal interpreted the LIBOR Definition to require a bank to always submit the lowest of the available rates in the range:

In the LIBOR Definition what is required is an assessment of the rate at which the panel bank “could borrow”. That must mean the cheapest rate at which it could borrow. A borrower “can” always borrow at a higher rate than the lowest on offer. But the higher rate would not reflect what the LIBOR benchmark is seeking to achieve, namely identification of the bank’s cost of borrowing in the wholesale cash market at the relevant moment of time. If in a stable and liquid market a submitting bank seeks and receives offers for a reasonable market size at the very time it is to make its submission, and receives offers ranging from 2.50% to 2.53%, it would accept the offer at 2.50%. It would be absurd to suggest that the LIBOR question could then properly be answered by a submission of 2.53%. The bank “could” borrow at that rate in the sense that it was a rate which was available, but that is obviously not what “could” means.

It is hard to imagine, in the abstract, a bank voluntarily borrowing at above the lowest rate then available. But finance is a complicated business, and borrowing decisions are not made in the abstract. There are plenty of examples where a bank might take a higher rate.

Crimes and contracts

Bear in mind that the “legal question” to be answered here is one of criminal law, not contract: whether the nebulous ingredients of common law ”conspiracy to defraud” were satisfied.

The LIBOR Definition was not a statute at all, let alone a criminal one. It formed part of a contract between the submitting banks and the BBA.

That is was not a criminal offence per se to fail to comply with LIBOR did not move the court:

That is not, however, determinative. It was not a criminal offence per se to fail to comply with the Take-over Code, but that did not stop it being treated in Spens as something which demanded construction as a question of law in the same way as primary or delegated legislation. Although compliance with LIBOR or EURIBOR was not directly a regulated activity, it was indirectly so: failure to comply with their provisions could give rise to regulatory consequences.

R v Spens concerned a breach of the Takeover Code, and expressed the view that “[the Takeover Code] sufficiently resembles legislation as to be likewise regarded as demanding construction of its provisions by a judge.” But inasmuch as the Takeover Code regulates the behaviour in connection with mergers and acquisitions it is a quasi-regulatory arrangement. The calculation of an interest rate benchmark is not. That failure to comply with its terms “could give rise to regulatory consequences” is beside the point: one may in breaching any contract also breach laws and regulations: this does not make the contract a quasi-regulatory arrangement requiring construction as a penal code.

For, unlike crimes and torts, contracts do not admit of mental states or “culpability”. There is no need for mens rea. You either comply with a contract, on the facts, or you don’t. One’s intention, recklessness or negligence does not come into it.

Furthermore, under the intellectual theory of criminal law, ignorance is no excuse. This is as axiomatic for an effective criminal justice system as “all interests in cash pass by delivery” is to finance: the criminal justice system would not work were defendants allowed to plead ignorance, even presumptively. Ignorantia legis non excusat, if you are blameless in your inadvertence, is a moral iniquity but still a practical imperative of good government.

But again, this does not hold for contract. Quite the opposite: under the intellectual theory of contract the parties are required to be materially cognisant of the whole thing. That is what offer and acceptance requires: if they are not present, there is no contract.

So the rules of contractual interpretation have forged a different path:

Interpretation is the ascertainment of the meaning which the document would convey to a reasonable person having all the background knowledge which would reasonably have been available to the parties in the situation in which they were at the time of the contract. [...] The background was famously referred to by Lord Wilberforce as the “matrix of fact,” but this phrase is, if anything, an understated description of what the background may include. Subject to the requirement that it should have been reasonably available to the parties and to the exception to be mentioned next, it includes absolutely anything which would have affected the way in which the language of the document would have been understood by a reasonable man.

—Lord Hoffman in Investors Compensation Scheme Ltd v West Bromwich Building Society [1998] 1 WLR 896

Interpreting the consensus ad idem manifested under a contract demands a wholly different approach than does construction of criminal legislation where the defendant’s understanding of the legislation is irrelevant.

Plainly, what a contract means is fact-dependent. A contract testifies to the parties’ mutual understanding and cannot, “mistakes of fact” notwithstanding, be sovereign to it.

Surely, evidence of how everyone behaved when interacting with the LIBOR Definition will help with what a reasonable person would have understood it to mean. There can be no better indication of reasonableness than direct evidence of the behaviour of fellow passengers on the Clapham Omnibus.

There is here the odd spectre of the law of contract forming the backdrop to a criminal allegation. This is rare. Usually, the criminal authorities stay well out of commercial disputes, even where allegations of fraud are flying around, seeing them as a matter of civil loss between merchants perfectly able to look after themselves, and not requiring the machinery of the state.

LIBOR, on which the bank deposits and mortgage repayments of unwitting retail punters depend, made things a bit different. This is no private matter to be sorted out between gentlemen with revolvers.

Nevertheless, still one must apply contractual principles, not criminal ones, to matters of contractual practice.

And the argument here is not about economic reality but legal meaning, and legal meaning follows natural, ordinary meaning, and in the world of contractual interpretation, that is viewed from the perspective of the person performing the contract, contra proferentem — against the draftsperson’s interest — giving the benefit of the doubt to the reader.

Defendants get the benefit of the same doubt in case of ambiguously framed crimes. For if the LIBOR Definition meant to mandate this “obvious” outcome, it did not do a very good job of it. As a matter of plain English, “could borrow” does not rule out a higher rate, but rather implies it: the Court of Appeal concedes as much, at para 89:

“The bank “could” borrow at that [higher] rate in the sense that it was a rate which was available, but that is obviously not what “could” means.”

The “obviousness” to which the Court appeals here is not a legal one — there is no authority for that proposition at all — but the Court’s economic intuition based upon an abstract conceptualisation of “borrowing”.

But borrowing does not happen in the abstract.

Per the plain words of the LIBOR Definition there is an upper bound delimited by the range of “inter-bank offers in reasonable market size just prior to 1100”. A submitter could not submit a rate higher than any actually offered, any more than it could submit a rate lower than one actually offered.

So, to construe “the rate at which it could borrow funds” to mean “the lowest rate ... ”, one must imply a term into the contract that is not there. Courts do not do this lightly. In Mackinnon LJ’s memorable words:

“That which in any contract is left to be implied and need not be expressed is something so obvious that it goes without saying; so that, if, while the parties were making their bargain, an officious bystander were to suggest some express provision for it in their agreement, they would testily suppress him with a common ‘Oh, of course!’”

There are good reasons to imagine at least thirty-seven LIBOR submitters might not have done this. Especially since it would have been easy enough for the old grandees to have put the matter beyond doubt, with a single modifying adjective:

“An individual BBA LIBOR Contributor Panel Bank will contribute the lowest rate at which it could borrow funds ...

They did not.

There were many other techniques they might have used to prevent banks talking their own book: for example, inviting them to submit the minimum rates they were prepared to lend to each other, rather than borrow.

They did not do that either.

As Hayes’ original pleading made clear, a bank submitting the rate at which it could borrow has an inherent conflict of interest. There were any number of ways it could craft the data it received in ways no-one could check: by carefully selecting the banks from whom it did, and did not, seek offers. From its timing. From the phrasing of the request.

If LIBOR had a problem it was not, principally, with the submitters: it was with the process. If you give merchants the flex to align their behaviour with their commercial interests, it is an odd merchant indeed who will not do it.

Everyone was at it

And after all, everyone was at it. A fun game, if you have twenty minutes, is to google the names of the LIBOR panel banks to see which were not somehow implicated in “LIBOR rigging”. If you haven’t got twenty minutes, the WSJ’s brilliant interactive spider network will give you the answer in an instant. There were thirty-seven prosecutions for LIBOR manipulation.

Everyone was at it.

We must draw one of two conclusions: either there was a colossal conspiracy by which everyone was trying to rip off the general public — and failing — or this is how everyone understood LIBOR to work.

It might not be edifying, but employees have fiduciary obligations to their shareholders, and if everyone acts according to those fiduciary obligations — or even their own personal self-interests — the selfishness cancels itself out. This is exactly the logic of Adam Smith’s invisible hand.

The inherent conflict of interest in marking your own homework

Indolent teachers at the JC’s school — there weren’t many other kinds — would sometimes tell pupils to mark their own homework. This never worked very well, unless the teacher hit upon the idea of asking the pupils to mark each other’s homework.

It is not clear what the theory underlying the LIBOR prosecutions was. We can speculate, but none survive close inspection.

Is the idea of a merchant prioritising its own commercial interests somehow reprehensible? This will be news to economists, and indeed the commercial courts who have frequently expected merchants to do nothing else.

Should LIBOR submitters should avoid conflicts of interest? How can they? The LIBOR rate is structurally fundamental to the economics of banking. All banks are exposed to interest rates. All have skin in the game. All are necessarily conflicted if asked to opine on what they think the interest rate should be. Any submission must, at some level, support or undermine the bank’s intrinsic interest rate exposure. A submission weighted to the lowest available rate structurally favours a bank that is not hedging its interest rate risk: why is that okay?

Is the LIBOR rate designed to protect investors, and if so who? Depositors? Borrowers? Why? As noted, classic bank customers, who borrow fixed and deposit floating, would benefit from a higher rate, not a lower one. This is of course, all very complicated, because banks are very complicated. It is not obvious what is or is not in a bank’s interest.

Stare decisis

This is a real lawyer nerd-out, but in forming its decision the Court of Appeal was confronted with some of its own prior rulings and judgments. The common law doctrine of precedent means an appeal court is generally bound by its own previous decisions in analogous cases. Usually, previous decisions are from unrelated cases. This always provides room for to distinguish inconvenient earlier decisions “on their facts”.

But not here. Here, the prior authorities were decided in previous appeals of the actual case before the Court of Appeal. This is unusual, due to the convoluted route by which the case came to the Court of Appeal, having been referred to it by the Criminal Cases Review Commission. Hayes and Palombo’s original convictions had already been appealed once to the Court of Appeal.

On one hand, it should not make a difference that it is the same case. It removes any possibility for “distinguishing on the facts”. On the other, asking the same court to reconsider decisions it feels constitutionally bound to follow makes a mockery of the appeal process. What is the point of reviewing a case you are bound to follow? This should, at least, be an unequivocal grounds for allowing a further appeal to the Supreme Court, which is not bound by lower court decisions or, after a famous practice direction, its own ones.

See also

• Interest rate swap mis-selling scandal
• Contract
• LIBOR